A comparative study regarding the adulteration detection of honey: Physicochemical parameters vs. impedimetric data

Honey adulteration is a major issue for European Union and its members because of an unfair practice of different producers and beekeepers, many adulterations involve the addition of sweet, concentrated syrups which may appear like honey. In our study we analysed the influence of adulteration of tilia honey with different syrups (such as corn, rice, inverted sugar, agave, maple syrups) in different percentages (5%, 10%, and 20% respectively) on physicochemical parameters (moisture content, L*, hab,cab, pH, free acidity, electrical conductivity (EC), 5-hydroxymetilfurfural (HMF), fructose, glucose, sucrose, turanose, trehalose, melesitose and raffinose) and impedimetric properties using electrochemical impedance spectroscopy. The impedimetric sensing was made using an electrochemical cell composed of two gold electrodes, and the frequency ranged between 0.1 kHz and 100 kHz. The impedimetric parameters (Z′, Z″ and phase) and Randal circuit parameters can distinguish the authentic honeys from the adulterated ones (based on the adulteration agent and adulteration percentage, respectively). The partial least squares – discriminant analysis (PLS-DA) and support vector machines (SVM) were used in a binary mode to separate the authentic honeys from the adulterated ones, and the SVM proved to separate much better than PLS-DA.


Introduction
Honey is the most natural sweetener used by human, being produced by bees from the nectar of plants or from sweet secretions of plants.From chemical point of view, it contains fructose, glucose, water, oligosaccharides, and small amounts of amino acids, minerals, vitamins, flavonoids, phenolic compounds, volatile compounds, organic acids (Escriche et al., 2011;Juan-Borrás et al., 2014a, 2014b;Oroian, 2012;Oroian et al., 2015a, b, c;Pauliuc et al., 2020;Pauliuc and Oroian, 2020).The European Union is the second largest producer of honey (275 000 tones) after China (500 000 tones); although UE is a large honey producer, yearly more than 170 000 tones are imported from non-UE countries (worth 405.9 million euros) such as: Ukraine, China, Argentina (Eurostat, 2021).Honey adulteration is a major issue for the authorities; the increasing of honey consumption led to an increase of fraudulent practices worldwide.The OPSON X (Operation OPSON is a Europol INTERPOL joint operation targeting fake and substandard food and beverages) operation ran from December 2020 to June 2021 and showed that 7% of the honey analysed were non-compliant products which means 51 000 kg of fraudulent treated honey.While in a study released by the EU named From the hives, 46% samples of a total of 320 samples analysed are suspicious and not complying with the UE directive on honey (From the Hives, 2023).In the USP database of adulterated food, honey is placed on the third position (Limm et al., 2023;Moore et al., 2012).In 2019, the Canadian government released a report which showed that 22% of 244 honey analysed were adulterated (Canadian Government Report: Enhanced Honey Authenticity Surveillance (2018to 2019), 2019), while in a survey conducted on Indian market revealed that all the honey were adulterated (Phipps, 2009).The main adulteration of honeys involves: addition of cheap sweeteners (e.g., sucrose from beet or canes, corn sugar, rice sugar, fructose, high-fructose corn syrups, maltose, inverted sugar), sugar feeding of bees or the mislabeling of the true origin of the product (botanical origin or geographical origin) (Cordella et al., 2002;El-Bialee and Sorour, 2011;Lawal Ridawan A. et al., 2009;Oroian et al., 2017Oroian et al., , 2019;;Piotr et al., 2016;Siddiqui et al., 2017aSiddiqui et al., , 2017b)).
Electrochemical impedance spectroscopy (EIS) is a rapid, simple, cost-effective, non-destructive and high-throughput electrochemical technique which was used for characterization of food products (T.K. Huang et al., 2021;Minetto et al., 2022).This technique consists in applying an alternating current or voltage signal (with a small sinusoidal perturbation at fixed frequency) to an electrochemical cell (e.g.2-4 electrodes) which contains the investigated matrix; the alternating current or voltage signal response at the same frequency is measured and transformed into impedance (T.K. Huang et al., 2021;Minetto et al., 2022;Soares et al., 2020).The Nyquist plot (where for each frequency, real part of impedance (Z') is plotted versus imaginary part of impedance (Z")) is obtained by applying the process described above to different frequencies (Aguedo et al., 2020;T. K. Huang et al., 2021).EIS was applied for the characterization of vegetable tissues (Ando et al., 2016) and beverages (Soares et al., 2020), for identifying of raw bovine milk adulteration with urea (Minetto et al., 2022), milk characterization (Lopes et al., 2018), determination of electrical parameters of honey (Paszkowski et al., 2014), determination of fish freshness (Sun et al., 2018).Therefore, this study aimed to evaluate the application of EIS to identify the intentional addition of different adulteration agents (corn, rice, inverted sugar, agave and maple syrups) to honey.To our knowledge this technique was not used for the detection of honey adulteration with the syrups studied in the present study.

Honey samples
22 samples of tilia honeys were purchased from local beekeepers from Suceava county, Romania, from the 2022 season.The samples were authenticated according to the melissopalynological method (Louveaux et al., 1978) (all samples had more than 60% of pollen grains from Tilia europaea).They were kept at 4 • C prior analysis.

Physicochemical parameters determination
The International Honey Commission methods were used for the determination of pH, free acidity, moisture content and electrical conductivity (Bogdanov, 2002).The color was determined in CIEL*a*b* coordinates using a CR-400 chromameter (Konica Minolta, Japan).The 5-hydroxymethylfurfural content was determined by the spectrophotometric method published by White (1979).High performance liquid chromatography (HPLC) was used to determine the sugar content (fructose, glucose, sucrose, turanose, trehalose, melesitose and raffinose) according to the method proposed by Bogdanov (2002).

EIS measurement
Two gold electrodes (2.0 mm in diameter) placed at 4 mm one to the other were introduced directly in the honey (20 g).The EIS measurement was carried out from 0.1 kHz to 100 kHz with an unbiased 100 mV sinusoidal signal using a PGSTAT 204 with FRA32M module (Metrohm, Filderstadt, Germany).The data were recorded and analysed with NOVA 2.0 software (Metrohm, Filderstadt, Germany).All analyzes were performed in triplicate, at 25 • C degree.
The EIS is a powerful technique for the characterization of electrochemical systems, and in particular on the characteristics of food products which contain certain substances dissolved (e.g.ions) which may conduct electricity.The impedance of system (Z) is calculated using the Ohm's law, and in Cartesian coordinates is given by: where Z′ is the real part of the impedance, Z″ is the imaginary part, and j = ̅̅̅̅̅̅ ̅ − 1 √ .

Statistical analysis
The results were submitted to analysis of variance (ANOVA) and Pearson correlation using XLSTAT trial version (Microsoft, Charlotte, NC, USA).Fisher's least significant difference (LSD) procedure was used at the 95% confidence level.
The partial least squares-discriminant analysis (PLS-DA) model and support vector machines (SVM) model were performed using XLSTAT trial version (Microsoft, USA).To attain precise classification of authentic and adulterated samples, various key parameters, including accuracy, sensitivity (which assesses the correct identification of true positives), and specificity (which assesses the correct identification of true negatives), were measured.These parameters were calculated using the following equations (Barbosa et al., 2016):

Sensitivity = True positives True positives + False negatives
(2) The data were split into two categories: calibration set (66.6% of the samples) and validation set (33.3% of the samples); the samples were chosen randomly by the software.

Physicochemical properties
In Table 1 are presented the physicochemical properties of tilia honeys analysed (mean, minimum, maximum, and standard deviations).The tilia honeys moisture content ranged between 17.0 and 19.9, below the limit established by the European Union at 20% (Eu Council, n.d.), the level was in agreement with the literature (Abera and Alemu, 2023;Oroian, 2015); a moisture content higher than 20% can depreciate the honey because the fermentation process which may occur.
The color is the main parameter which influences the perception of consumer toward a certain food, and it is described by: lightness L* (white/dark), h* ab (huerepresents the perceived color of an object) and C* ab (chromadescribes the intensity, saturation or purity of a color, high values belonging to rich colors and low values to grey colors) (Tuberoso et al., 2014).As Table 1 shows, the L* parameter value ranged between 34.3 and 42.3 with a mean value of 38.2, the h ab ranged between 87.6 and 92.2, while the C* ab ranged between 26.7 and 35.7.The values were in agreement with other studies made on tilia honeys (Juan-Borrás et al., 2014a;Pauliuc et al., 2021).
The electrical conductivity is influenced by the level of minerals dissolved in and by the presence of organic acids in honeys (Kivrak et al., 2017a, b).The electrical conductivity ranged between 337.0 and 763.00 μS cm − 1 (Table 1) values which were closed to those reported honey tilia honeys from Spain, Czech Republic, and Italy (Juan-Borrás et al., 2014a;Persano Oddo and Piro, 2004).
The pH and the free acidity of honey are correlated with the extraction and storage of it, and the level of them may influence the shelf life of the product (Kivrak et al., 2017b).The pH values ranged between 4.1 and 5.6 (Table 1), while the free acidity ranged between 1.0 and 12.6 meq acid/100 g, which is below the level set by UE (Eu Council, n.d.).
Hydroxymethylfurfural (HMF) is a quality criterion for honey and occurs due to the chemical reactions which may be formed when the product is kept at unsuitable temperature or the heat treatment is inadequate (Shapla et al., 2018).The HMF ranged between 0.4 and 26.1 mg/kg honey (Table 1), below the level of 40 mg/kg established for honey (Eu Council, n.d.), the values of the HMF are in the agreement with the literature (Besir et al., 2021;Grainger et al., 2017Grainger et al., , 2017n, 2017ret al., 2018)).
The individual sugar profile of each honey may be used for the botanical authentication, and they have an impact on the physicochemical properties (e.g., osmolarity), and the fructose/glucose ratio influence the crystallization process (Islam et al., 2022).The main sugar presented is fructose (31.9-35.1%),followed by glucose (26.8-32.0%)and sucrose (0.1-2.2%).The ratio fructose/glucose (F/G) is between 1.1 and 1.2, and for this reason it has a crystallize tendency shortly after extraction.The melesitose ranged between 0.0 and 2.4%.Turanose, trehalose, and raffinose were below the limit of quantification.All the data can be found in Table 1.

Influence of adulteration agents on physicochemical parameters
In Table 2 are presented the addition of different percentages (5, 10 and 20%) of sugars syrups into honey (maple, agave, rice, corn, and inverted sugars).The L*, fructose and glucose content decreased significantly with the addition of maple syrup addition into honey, the electrical conductivity and sucrose content increased significantly with the addition of the maple syrup.The moisture content was not influenced significantly by the addition of maple syrup but at concentration higher than 10%, the level was higher than 20% (the threshold established by the UE).The electrical conductivity, glucose content, decreased significantly with the addition of agave syrup, while HMF, fructose content, F/G increased significantly.L*, the fructose content, glucose content, sucrose content and F/G decreased significantly with the addition of rice syrup, while the free acidity and melesitose increased significantly, respectively.EC, fructose content, glucose content decreased significantly with the addition of corn syrup.The inverted sugar addition influenced the EC which decreased significantly, while the HMF increased significantly, the addition of more than 10% syrup lead to a concentration higher than the level established by the UE (the inverted sugar has a high concentration of HMF because of the heat treatment applied during the sucrose conversion (Ciursa and Oroian, 2021a)).From all the 5 syrups used for the adulteration, the maple syrup generated the most intensive changes (e.g.L*, EC, glucose, sucrose).

Honey EIS properties
The projection of the real part of the impedance vs. the imaginary part of the impedance is called Nyquist plot.In Fig. 1 is presented a typical Nyquist plot for tilia honey samples (the frequency ranged between 0.1 KHz and 100 kHz); the electrochemical active species presented into honeys are phenols, vitamins are prone to generate an exchange current at electrode interface by means of a small potential.Honey contains a large spectrum of organic acids (e.g., gluconic, formic, lactic, acetic, citric, malic acids) (Pauliuc and Oroian, 2020); these compounds can act as functioning as supporting electrolytes which may diminish the Ohmic drop phenomenon and inhibit the migration of electroactive substances toward the polarized electrodes, ultimately leading to an augmentation in electrochemical current density (Tsierkezos and Ritter, 2012).The differences between the Nyquist plot of the tilia sample is because of the electrical conductivity of the samples (EC = 537-763 μS/cm), which influence the electrical circuit.In Fig. 2 is presented the Nyquist plot for the honey (the sample with medium percentage of Tilia europaea pollen grains -65.2%) adulterated with agave, inverted sugar, maple, corn, and rice syrups in different concentrations (5%, 10% and 20%).As can be observed in Fig. 2 the addition of adulteration agents led to a decrease of the circle radius of Nyquist plot.To achieve a fitting of the data, the experimental impedance spectra data were subjected to a conventional Randles equivalent circuit model (depicted in Fig. 3), encompassing components such as solution resistance (R S ), double layer capacitance (CD) and charge transfer resistance (R CT ).In Table 3 are presented the data about the impedance, phase and Randles model parameters; the data are presented in two ways: the first one authentic vs. adulterated samples, and the second one authentic vs. each adulteration agent and its addition percentage.When we compare the authentic honeys vs. adulterated ones, there can be observed high differences with respect to Z′, Z″, phase, R s and R 1.The solution resistances of the adulterated are much higher because the content of electrochemically active species and supporting electrolytes are much lower (T.K. Huang et al., 2021).In the case of each adulteration agent, there can be observed that in the case of maple syrups addition, the increase of the adulteration agent led to a decrease of the R s in opposition with the behaviour of the other adulterations, this fact maybe attributed to the electrochemical species presented (the electrical conductivity of the samples with maple syrups are much higher (Table 2)).The Warburg impedance decreased with the increasing of the adulteration agents.The R s , CD and R CT of the samples analysed are different from the literature because the electrical conductivity of the authentic honey taken for analysis were much lower (T.K. Huang et al., 2021).
The group comparisons among authentic honeys and adulterated honeys were made using ANOVA with Fisher's least significant difference (LSD) procedure at 95% confidence level; there were observed a significant difference (P < 0.05) between the parameter's studies presented in Table 3 (double layer capacitance have not a clear evolution with the adulteration agent addition).The addition of 5% adulteration agents (agave, inverted sugar, rice and corn sugars) has significantly influenced the parameters presented in Table 3.

Honey adulteration detection
Honey adulteration detection involved the utilization of two statistical methodologies: partial least squaresdiscriminant analysis (PLS-DA) and support vector machines (SVM).The two models were applied to: (a) physicochemical data (the parameters from Table 1), (b) impedimetric data (Z′, Z″ and phase) and (c) physicochemical + impedimetric data, in order to discriminate the authentic honeys from the adulteration ones, and also the classification of authentic honeys from each adulteration type (agave, maple, corn, inverted sugar, and rice sugars).

Honey adulteration detection -classification of pure honey and adulterated honeys using PLS-DA and SVM
The PLS-DA and SVM models were checked to see which of the two statistical analyses can better distinguish the samples using the physicochemical data, impedimetric data and physicochemical + impedimetric data.It was taken into consideration the accuracy, sensitivity and specificity as indicators for selecting the best analyses.In Table 4 are presented the statistical analysis results for PLS-DA and SVM.
The PLS-DA models reached lower parameters for the discrimination of the samples (authentic vs. adulterated) than the SVM models; in some cases, the accuracy, sensitivity and specificity of the SVM models were 100%, much higher than in the PLS-DA models.
Based on the input data, the PLS-DA models built on physicochemical data had statistical parameters much better than on the models built on impedimetric data, while the models built on physicochemical + impedimetric data had the best statistical parameters.The SVM models built using the impedimetric data yielded superior statistical outcomes compared models built with physicochemical data.
The adulteration detection in this part was focused on the classification of the authentic honeys from the adulterated ones in function of the adulteration percentage (0% vs 5%, 0% vs 10%, 0% vs 20%).The

Table 2
Physicochemical parameters variation in function of adulteration with maple, agave, rice, corn, and inverted syrups.

Table 3
Effect of adulteration on the honey impedance properties.

Table 4
Binary classification of pure honey and adulterated honeys using PLS-DA and SVM based on (a) physicochemical data, (b) impedimetric data and (c) physicochemical + impedimetric data.